A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Effect of Machine Learning Algorithms on Detection of Landslides Caused by the 2015 Lefkada Earthquake
Machine learning algorithms can be used to detect landslides and develop landslide inventories following natural disasters. This study aims to evaluate the performance of different machine learning algorithms, specifically supervised methods, in detecting landslides on a complete high-quality landslide inventory. The Mw6.5 Lefkada earthquake that occurred on November 17, 2015, causing more than 700 landslides, was selected as the case study. The classification model used Worldview satellite imagery and digital elevation models (DEM) as input. Three machine learning algorithms were considered: random forest (RF), support vectors machine (SVM), and maximum likelihood (ML). All three machine learning algorithms classify landslides generally successfully, and RF achieves the best performance. Amalgamation is a typical issue in landslide detection and mapping, where multiple adjacent landslides are classified and mapped as one. This paper used flow direction to separate amalgamated landslides in the detection result. The result shows that the RF classifier captures large landslides more accurately than small landslides. Satellite imagery with higher resolution results in better classification performance than coarse-resolution imagery.
Effect of Machine Learning Algorithms on Detection of Landslides Caused by the 2015 Lefkada Earthquake
Machine learning algorithms can be used to detect landslides and develop landslide inventories following natural disasters. This study aims to evaluate the performance of different machine learning algorithms, specifically supervised methods, in detecting landslides on a complete high-quality landslide inventory. The Mw6.5 Lefkada earthquake that occurred on November 17, 2015, causing more than 700 landslides, was selected as the case study. The classification model used Worldview satellite imagery and digital elevation models (DEM) as input. Three machine learning algorithms were considered: random forest (RF), support vectors machine (SVM), and maximum likelihood (ML). All three machine learning algorithms classify landslides generally successfully, and RF achieves the best performance. Amalgamation is a typical issue in landslide detection and mapping, where multiple adjacent landslides are classified and mapped as one. This paper used flow direction to separate amalgamated landslides in the detection result. The result shows that the RF classifier captures large landslides more accurately than small landslides. Satellite imagery with higher resolution results in better classification performance than coarse-resolution imagery.
Effect of Machine Learning Algorithms on Detection of Landslides Caused by the 2015 Lefkada Earthquake
Huang, Jhih-Rou (author) / Zekkos, Dimitrios (author)
Geo-Congress 2023 ; 2023 ; Los Angeles, California
Geo-Congress 2023 ; 347-356
2023-03-23
Conference paper
Electronic Resource
English
| British Library Conference Proceedings | 2023
| British Library Conference Proceedings | 2021
The 2003 LEFKADA (Greece) Earthquake OSP 13
| British Library Conference Proceedings | 2004